Deep Learning-Based Method and Device for Monitoring Transformer Temperature Distribution

Abstract

This invention discloses a method and related apparatus for monitoring transformer temperature distribution based on deep learning. The monitoring method includes: acquiring the current voltage and current data of the transformer; using the current voltage and current data and a pre-trained deep learning model to predict the temperature distribution, obtaining a current prediction result for the transformer, including at least the current winding short-circuit location, the current risk location, and the current temperature distribution. The risk location indicates the location where the transformer has a risk of winding burnout. Through this method, the acquired voltage and current data can be input into the deep learning model to predict the transformer's temperature distribution, winding short-circuit location, and risk location with the risk of winding burnout. This achieves transformer temperature distribution monitoring and locates short-circuit and risk locations to assess the transformer's health status without the need for temperature sensors.

Description

Technical Field

[0001] This invention relates to the field of data monitoring technology, and in particular to a method and related apparatus for monitoring transformer temperature distribution based on deep learning. Background Technology

[0002] Transformers are common equipment in power systems, used to transform voltage and current transmission. As the core component of a transformer, the windings are subjected to the forces of current and magnetic fields, causing stress and deformation. This can lead to short-circuit faults within the transformer or between windings. Short-circuit faults in transformers can cause temperatures at certain locations to exceed the maximum permissible temperature, leading to risks of insulation breakdown or winding overheating and burnout, and even potentially causing fires, explosions, and other safety accidents that threaten human lives. Early detection of abnormal temperatures within the transformer can reduce potential fault risks and improve the stability of the power system. Currently, monitoring transformer temperature distribution mainly involves directly measuring the temperature at certain locations using temperature sensors. However, for some locations within the transformer's internal windings, which are sealed and surrounded by the core and other windings, it is impossible to install temperature sensors for direct temperature measurement, affecting the assessment of the transformer's overall health status.

[0003] Therefore, a reasonable and effective method is needed to obtain detailed information on transformer temperature distribution using simple and measurable transformer experimental data, thereby providing early warning of transformer health status. Summary of the Invention

[0004] The main objective of this invention is to provide a method and related device for monitoring transformer temperature distribution based on deep learning, which can solve the problem in the prior art that it is impossible to install temperature sensors to directly measure temperature.

[0005] To achieve the above objectives, the first aspect of the present invention provides a method for monitoring transformer temperature distribution based on deep learning, the monitoring method comprising:

[0006] Obtain the current voltage and current data of the transformer;

[0007] Temperature distribution is predicted using the current voltage and current data and a pre-trained deep learning model to obtain the current prediction result of the transformer. The current prediction result includes at least the current winding short-circuit position, the current risk position, and the current temperature distribution of the transformer. The risk position is used to indicate the location where the transformer has a risk of winding burnout. The deep learning model is trained using a pre-collected training sample set, which includes at least the first risk position, the first voltage and current data, and the first temperature distribution under each winding short-circuit position of the transformer.

[0008] In one feasible implementation, the method further includes:

[0009] Construct a finite element model of the transformer;

[0010] An initial simulation is performed based on the finite element model to obtain the initial characteristic parameters of the finite element model, wherein the initial characteristic parameters include at least the maximum temperature parameter;

[0011] Using each preset winding short-circuit position as an independent variable, the finite element model is used to simulate the winding short-circuit condition, and the first voltage and current data and the first temperature distribution under each winding short-circuit position are obtained.

[0012] Based on the comparison between the maximum temperature parameter and the first temperature distribution, the first risk location under each winding short-circuit location is obtained.

[0013] In one feasible implementation, the step of comparing the maximum temperature parameter with the first temperature distribution to obtain the first risk location under each winding short-circuit location includes:

[0014] Each temperature data in the first temperature distribution is compared with the temperature threshold corresponding to the maximum temperature parameter to obtain the comparison result;

[0015] If the comparison result is that the temperature data is greater than the temperature threshold, then it is determined that the transformer location corresponding to the temperature data is at risk of burning out, and is marked as the first risk location.

[0016] In one feasible implementation, constructing the finite element model of the transformer includes:

[0017] The finite element geometric model of the transformer is divided into the core, windings, and oil tank.

[0018] The materials of the core, windings and tank are set according to the actual transformer parameters to obtain the set finite element geometric model;

[0019] By setting the corresponding boundary conditions for the core and windings in the finite element geometric model, the finite element model of the transformer is obtained.

[0020] In one feasible implementation, the initial simulation based on the finite element model to obtain the initial characteristic parameters of the finite element model includes:

[0021] The finite element model is subjected to excitation to perform initial simulation of transient and steady-state simulation, the second current and voltage data of the transformer are obtained, and the maximum temperature parameter of the transformer during the transient and steady-state simulation is calculated.

[0022] In one feasible implementation, the step of using each preset winding short-circuit position as an independent variable to simulate the short-circuit condition of the finite element model, and obtaining the first voltage and current data and the first temperature distribution at each winding short-circuit position, includes:

[0023] The short-circuit positions of the windings are traversed sequentially. For each target short-circuit position, the short-circuit condition of the transformer is set as the target short-circuit position to simulate the short-circuit condition. The first current and voltage data under the target short-circuit position are obtained, and the first temperature distribution of the transformer under the short-circuit condition at the target short-circuit position is calculated.

[0024] To achieve the above objectives, a second aspect of the present invention provides a deep learning-based transformer temperature distribution monitoring device, the device comprising:

[0025] Data acquisition module: used to acquire the current voltage and current data of the transformer;

[0026] Temperature prediction module: used to predict temperature distribution using the current voltage and current data and a pre-trained deep learning model to obtain the current prediction result of the transformer. The current prediction result includes at least the current winding short-circuit position, the current risk position, and the current temperature distribution of the transformer. The risk position is used to indicate the location where the transformer has a risk of winding burnout. The deep learning model is trained using a pre-collected training sample set. The training sample set includes at least the first risk position, the first voltage and current data, and the first temperature distribution under each winding short-circuit position of the transformer.

[0027] In one feasible implementation, the apparatus further includes:

[0028] Model building module: used to construct the finite element model of the transformer;

[0029] Parameter determination module: used to perform initial simulation based on the finite element model to obtain the initial characteristic parameters of the finite element model, wherein the initial characteristic parameters include at least the maximum temperature parameter;

[0030] Winding short-circuit simulation module: used to simulate the short-circuit condition of the finite element model with each preset winding short-circuit position as the independent variable, and obtain the first voltage and current data and the first temperature distribution under each winding short-circuit position.

[0031] Risk location module: used to compare the maximum temperature parameter and the first temperature distribution to obtain the first risk location under each winding short circuit position.

[0032] To achieve the above objectives, a third aspect of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps shown in the first aspect and any feasible implementation.

[0033] To achieve the above objectives, a fourth aspect of the present invention provides a computer device including a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps shown in the first aspect and any feasible implementation.

[0034] The embodiments of the present invention have the following beneficial effects:

[0035] This invention provides a deep learning-based method for monitoring transformer temperature distribution. The method includes: acquiring the current voltage and current data of the transformer; using the current voltage and current data and a pre-trained deep learning model to predict the temperature distribution, obtaining the current prediction result of the transformer. The current prediction result includes at least the current winding short-circuit location, the current risk location, and the current temperature distribution of the transformer. The risk location indicates the location where the transformer has a risk of winding burnout. The deep learning model is trained using a pre-collected training sample set, which includes at least the first risk location, first voltage and current data, and first temperature distribution under each winding short-circuit location of the transformer. Through this method, the collected voltage and current data can be input into the deep learning model to predict the transformer's temperature distribution, winding short-circuit location, and risk location with a risk of winding burnout. This achieves transformer temperature distribution monitoring and locates short-circuit and risk locations to determine the transformer's health status without the need for temperature sensors. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] in:

[0038] Figure 1 This is a flowchart of a transformer temperature distribution monitoring method based on deep learning, as described in an embodiment of the present invention.

[0039] Figure 2 This is another flowchart of a deep learning-based method for monitoring transformer temperature distribution in an embodiment of the present invention;

[0040] Figure 3This is a structural block diagram of a transformer temperature distribution monitoring device based on deep learning, according to an embodiment of the present invention.

[0041] Figure 4 This is a structural block diagram of a computer device in an embodiment of the present invention. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] Please see Figure 1 , Figure 1 This is a flowchart illustrating a deep learning-based method for monitoring transformer temperature distribution in an embodiment of the present invention. This method can be applied to either a terminal or a server. The terminal can be a desktop terminal or a mobile terminal; a mobile terminal can be at least one of a mobile phone, tablet, or laptop. The server can be a standalone server or a server cluster composed of multiple servers. This embodiment uses a terminal application as an example. Figure 1 The methods shown include:

[0044] 101. Obtain the current voltage and current data of the transformer;

[0045] 102. Using the current voltage and current data and a pre-trained deep learning model, temperature distribution is predicted to obtain the current prediction result of the transformer. The current prediction result includes at least the current winding short-circuit position, the current risk position, and the current temperature distribution of the transformer. The risk position is used to indicate the position where the transformer has a risk of winding burnout. The deep learning model is trained using a pre-collected training sample set. The training sample set includes at least the first risk position, the first voltage and current data, and the first temperature distribution under each winding short-circuit position of the transformer.

[0046] It should be noted that artificial intelligence databases are models obtained through training and deep learning on large amounts of data, possessing strong information processing and judgment capabilities. Therefore, this application provides a deep learning-based method for monitoring transformer temperature distribution. During actual transformer operation, the transformer temperature distribution is predicted using easily obtainable and easily collected transformer port current and voltage waveform parameters. Specifically, the initial deep learning model is trained using a pre-collected training sample set. Since this training sample set includes at least the first risk location, first voltage and current data, and first temperature distribution under each winding short-circuit position of the transformer, that is, the location of the transformer with winding burnout risk under each winding short-circuit position, port voltage and current data, and temperature distribution, the deep learning model trained in this way can learn the transformer port current and voltage waveform data, temperature distribution, and the risk location under short-circuit conditions, thereby achieving temperature distribution monitoring.

[0047] The training sample set can be obtained through simulation analysis. By simulating different operating conditions, corresponding data performance is obtained, thereby constructing a database of training sample sets for deep learning models to learn the relationship between different feature data. In this way, the performance of other data can be predicted using one data that can be actually collected, such as predicting temperature distribution and short circuit location using voltage and current data.

[0048] Transformer temperature distribution monitoring is achieved using deep learning. In actual operation, transformer port current and voltage waveform parameters are readily available. Using deep learning technology, the obtained port current and voltage waveforms are compared point-by-point with those in a database according to time sequence to determine potential short-circuit locations, the current temperature distribution, and locations at risk of winding overheating and burnout.

[0049] This invention provides a deep learning-based method for monitoring transformer temperature distribution. The method includes: acquiring the current voltage and current data of the transformer; using the current voltage and current data and a pre-trained deep learning model to predict the temperature distribution, obtaining the current prediction result of the transformer. The current prediction result includes at least the current winding short-circuit location, the current risk location, and the current temperature distribution of the transformer. The risk location indicates the location where the transformer has a risk of winding burnout. The deep learning model is trained using a pre-collected training sample set, which includes at least the first risk location, first voltage and current data, and first temperature distribution under each winding short-circuit location of the transformer. Through this method, the collected voltage and current data can be input into the deep learning model to predict the transformer's temperature distribution, winding short-circuit location, and risk location with a risk of winding burnout. This achieves transformer temperature distribution monitoring and locates short-circuit and risk locations to determine the transformer's health status without the need for temperature sensors.

[0050] Please see Figure 2 , Figure 2 This is another flowchart of a deep learning-based transformer temperature distribution monitoring method according to an embodiment of the present invention, as shown below. Figure 2 The method shown includes the following steps:

[0051] 201. Construct the finite element model of the transformer;

[0052] It is understandable that in order to complete the training of a deep learning model, it is necessary to collect a training sample set in advance. In order to obtain more accurate and reliable training samples, this application constructs a database through simulation to collect sample data. First, in order to realize the simulation of the transformer, it is necessary to construct a simulation model of the transformer. This simulation model can be a finite element model, which is made according to the actual transformer data.

[0053] In one feasible implementation, step 201 includes: dividing the finite element geometric model of the transformer into core, winding and tank; setting the materials of the core, winding and tank according to the actual transformer parameters of the transformer to obtain the set finite element geometric model; setting the corresponding boundary conditions for the core and winding in the set finite element geometric model to obtain the finite element model of the transformer.

[0054] For example, a model can be built in COMSOL finite element simulation software based on actual transformer parameters. Specifically, the finite element geometric model of the transformer is divided into three parts: the core, the winding, and the tank. Then, the materials of each geometric part are set according to the actual transformer, and the corresponding boundary conditions are set for the core and the winding to obtain the finite element model of the transformer.

[0055] 202. Perform initial simulation based on the finite element model to obtain the initial characteristic parameters of the finite element model, wherein the initial characteristic parameters include at least the maximum temperature parameter;

[0056] Furthermore, in order to obtain the initial characteristic parameters of the finite element model and perform initial simulation, the transient and steady-state simulations are achieved by applying excitation to the finite element model to obtain the voltage and current data and temperature distribution at this time. In order to identify the risk location later, the maximum temperature parameter is also calculated at this time. The maximum temperature parameter is used to reflect the maximum temperature that the transformer can withstand to avoid overheating.

[0057] In one feasible implementation, step 202 includes: applying excitation to the finite element model to perform an initial simulation of transient and steady-state simulation, obtaining the second current and voltage data of the transformer, and calculating the maximum temperature parameter of the transformer during the transient and steady-state simulation.

[0058] For example, the specific model characteristic parameters obtained by simulation can be: applying excitation to the finite element model of the transformer, conducting transient and steady-state simulation studies, obtaining port current and voltage waveform data, and calculating the maximum temperature parameters of the transformer during operation based on the calculation function of the simulation software.

[0059] 203. Using each preset winding short-circuit position as the independent variable, the finite element model is simulated under the winding short-circuit condition to obtain the first voltage and current data and the first temperature distribution under each winding short-circuit position.

[0060] Furthermore, in order to obtain data performance under different short-circuit conditions, the above finite element model is simulated under different conditions. Specifically, the short-circuit positions of each winding are preset, and each preset short-circuit position is used as an independent variable to simulate the winding short-circuit conditions of the finite element model, so as to obtain the first voltage and current data and the first temperature distribution under each winding short-circuit position.

[0061] In one feasible implementation, step 203 includes: sequentially traversing the short-circuit positions of the windings; for each target short-circuit position traversed, setting the short-circuit condition of the transformer to the target short-circuit position and performing a short-circuit condition simulation to obtain the first current and voltage data under the target short-circuit position, and calculating the first temperature distribution of the transformer during operation under the short-circuit condition at the target short-circuit position.

[0062] For example, a .m file is written to implement parametric modeling and traversal simulation: In COMSOL, the initial simulation file obtained based on the initial simulation steps is exported as a .m file adapted for MATLAB. Then, the initial .m file is rewritten using MATLAB, with the winding short-circuit position as the independent variable. The automatic change of the winding short-circuit position is implemented through the code of loop statements. The function of each loop process is: to set the transformer winding short-circuit position, to obtain the port current and voltage waveform data, to calculate the temperature distribution of the transformer under this operating condition, and finally to store the port current, voltage and temperature distribution parameters in a specified file table.

[0063] 204. Based on the maximum temperature parameter and the first temperature distribution, the first risk location under each winding short-circuit location is obtained;

[0064] After obtaining the temperature distribution under short-circuit conditions, the risk location can be further identified. When the temperature exceeds the maximum temperature, it is considered that there is a risk of burnout. The risk location can be located by comparing the maximum temperature parameter with the first temperature distribution.

[0065] In one feasible implementation, step 204 includes:

[0066] Each temperature data in the first temperature distribution is compared with the temperature threshold corresponding to the maximum temperature parameter to obtain a comparison result; if the comparison result is that the temperature data is greater than the temperature threshold, it is determined that the transformer location corresponding to the temperature data has a risk of burnout and is marked as the first risk location.

[0067] The maximum temperature parameter is taken as 80% of the maximum temperature as the temperature threshold corresponding to the maximum temperature. By comparing and locating transformers with temperatures greater than this temperature threshold, risk locations are marked.

[0068] For example, a temperature monitoring database adapted to the transformer is constructed. The .m file obtained in the above steps is imported into COMSOL software, and simulation results for all winding short-circuit locations are obtained. The obtained data includes the short-circuit locations and their corresponding port current and voltage waveform data and temperature distribution. If the temperature parameter at a certain location is greater than 80% of the maximum allowable temperature at that location, it is considered that there is a risk of winding overheating and burnout at that location, and it is marked in the corresponding data. A temperature monitoring database adapted to the transformer is constructed, and the database of training sample datasets is obtained in sequence.

[0069] Furthermore, short-circuit location, risk location, and temperature distribution can be used as true labels for voltage and current data. This voltage and current data is then input into an initial deep learning model for prediction, yielding a first prediction result. The loss value between this first prediction result and the true labels of the input voltage and current data is used to determine if the model has converged. If the model converges, training is considered complete, and true labels can be predicted based on the input current and voltage data. Finally, the trained deep learning model is used to predict actual voltage and current.

[0070] 205. Obtain the current voltage and current data of the transformer;

[0071] 206. Using the current voltage and current data and a pre-trained deep learning model, a temperature distribution prediction is performed to obtain the current prediction result for the transformer. The current prediction result includes at least the current winding short-circuit location, the current risk location, and the current temperature distribution of the transformer. The risk location indicates the location where the transformer has a risk of winding burnout. The deep learning model is trained using a pre-collected training sample set, which includes at least the first risk location, first voltage and current data, and first temperature distribution under each winding short-circuit location of the transformer.

[0072] It should be noted that steps 205 and 206 are related to... Figure 1 Steps 101 and 102 shown are similar and will not be repeated here to avoid repetition. Please refer to [link to relevant documentation] for details. Figure 1 The contents of steps 101 and 102 are shown.

[0073] This invention proposes a deep learning-based method for monitoring transformer temperature distribution. This method combines COMSOL finite element simulation software with MATLAB to perform parametric modeling of the transformer, traversing different short-circuit locations, and simulating and analyzing the temperature field distribution under various operating conditions. Finally, a database is constructed using the simulation results, and deep learning is used to analyze the online measured port data and database parameters to achieve transformer temperature distribution monitoring. Specifically, the method includes the following processing steps S1 to S5:

[0074] S1: Building a model in COMSOL finite element simulation software based on actual transformer parameters: The finite element geometric model of the transformer is divided into three parts: the core, the winding, and the tank. Then, the materials of each geometric part are set according to the actual transformer. Finally, the corresponding boundary conditions are set for the core and the winding to obtain the finite element model of the transformer.

[0075] S2: Simulation to obtain model characteristic parameters: Excitation is applied to the finite element model of the transformer, and transient and steady-state simulation studies are conducted to obtain port current and voltage waveform data. Based on the software's calculation function, the maximum temperature parameter during transformer operation is calculated.

[0076] S3: Write the .m file for parametric modeling and traversal simulation: In COMSOL, export the initial simulation file obtained based on step S2 as a .m file adapted for MATLAB. Then, rewrite the initial .m file using MATLAB, using the winding short-circuit position as the independent variable. Implement automatic changes to the winding short-circuit position through loop code. Each loop process performs the following functions: sets the transformer winding short-circuit position, acquires port current and voltage waveform data, calculates the temperature distribution of the transformer under this operating condition, and finally stores the port current, voltage, and temperature distribution parameters in a designated file table.

[0077] S4: Construct a temperature monitoring database adapted to this transformer. Import the .m file obtained in the above steps into COMSOL software and simulate all winding short-circuit locations to obtain simulation results. The traversed data includes each short-circuit location and its corresponding port current and voltage waveform data and temperature distribution. If the temperature parameter at a certain location is greater than 80% of the maximum allowable temperature at that location, it is considered that there is a risk of winding overheating and burnout at that location, and this is marked in the corresponding data. Construct a temperature monitoring database adapted to this transformer.

[0078] S5: Transformer temperature distribution monitoring based on deep learning. In actual operation, transformer port current and voltage waveform parameters are readily available. Based on deep learning technology, the obtained port current and voltage waveforms are compared point-by-point with current and voltage waveforms in the database according to time sequence to determine the possible short-circuit locations of the transformer, the current temperature distribution of the transformer, and the locations at risk of winding overheating and burnout.

[0079] The advantage of this invention lies in its ability to utilize readily available transformer port current and voltage waveform parameters, based on deep learning, to obtain detailed transformer temperature distribution information, thereby providing early warning of transformer health status. This is of great significance for the safe operation, fault analysis, and prevention of transformers.

[0080] In one feasible implementation... Figure 3 This is a structural block diagram of a transformer temperature distribution monitoring device based on deep learning, as described in an embodiment of the present invention. Figure 3 The apparatus shown includes:

[0081] Data acquisition module 301: used to acquire the current voltage and current data of the transformer;

[0082] Temperature prediction module 302: used to predict the temperature distribution using the current voltage and current data and a pre-trained deep learning model, to obtain the current prediction result of the transformer. The current prediction result includes at least the current winding short-circuit position, the current risk position, and the current temperature distribution of the transformer. The risk position is used to indicate the location where the transformer has a risk of winding burnout. The deep learning model is trained using a pre-collected training sample set. The training sample set includes at least the first risk position, the first voltage and current data, and the first temperature distribution under each winding short-circuit position of the transformer.

[0083] It should be noted that, Figure 3 The function of each module in the device shown is as follows: Figure 1 The steps in the method shown are similar, and will not be repeated here to avoid repetition. For details, please refer to [reference needed]. Figure 1 The content of each step in the method shown.

[0084] This invention provides a deep learning-based transformer temperature distribution monitoring device. The device includes: a data acquisition module for acquiring the current voltage and current data of the transformer; and a temperature prediction module for using the current voltage and current data and a pre-trained deep learning model to predict the temperature distribution and obtain a current prediction result for the transformer. The current prediction result includes at least the current winding short-circuit location, the current risk location, and the current temperature distribution of the transformer. The risk location indicates the location where the transformer has a risk of winding burnout. The deep learning model is trained using a pre-collected training sample set, which includes at least the first risk location, first voltage and current data, and first temperature distribution under each winding short-circuit location of the transformer. By using the acquired voltage and current data as input to the deep learning model, the transformer's temperature distribution, winding short-circuit location, and risk location with a risk of winding burnout can be predicted. This achieves transformer temperature distribution monitoring and locates short-circuit and risk locations to assess the transformer's health status without the need for temperature sensors.

[0085] In one feasible implementation, the apparatus further includes:

[0086] Model building module: used to construct the finite element model of the transformer;

[0087] Parameter determination module: used to perform initial simulation based on the finite element model to obtain the initial characteristic parameters of the finite element model, wherein the initial characteristic parameters include at least the maximum temperature parameter;

[0088] Winding short-circuit simulation module: used to simulate the short-circuit condition of the finite element model with each preset winding short-circuit position as the independent variable, and obtain the first voltage and current data and the first temperature distribution under each winding short-circuit position.

[0089] Risk location module: used to compare the maximum temperature parameter and the first temperature distribution to obtain the first risk location under each winding short circuit position.

[0090] It should be noted that the functions of the aforementioned model building module, parameter determination module, winding short-circuit simulation module, and risk location module are the same as... Figure 2 The steps in the method shown are similar, and will not be repeated here to avoid repetition. For details, please refer to [reference needed]. Figure 2 The content of each step in the method shown.

[0091] The advantage of this invention lies in its ability to utilize readily available transformer port current and voltage waveform parameters, based on deep learning, to obtain detailed transformer temperature distribution information, thereby providing early warning of transformer health status. This is of great significance for the safe operation, fault analysis, and prevention of transformers.

[0092] Figure 4 An internal structural diagram of a computer device in one embodiment is shown. This computer device can specifically be a terminal or a server. Figure 4 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and may also store a computer program, which, when executed by the processor, causes the processor to perform the aforementioned methods. The internal memory may also store a computer program, which, when executed by the processor, causes the processor to perform the aforementioned methods. Those skilled in the art will understand that… Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0093] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform actions such as... Figure 1 or Figure 2 The steps of the method shown.

[0094] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, causes the processor to perform the following actions: Figure 1 or Figure 2 The steps of the method shown.

[0095] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0096] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0097] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for monitoring transformer temperature distribution based on deep learning, characterized in that, The monitoring method includes: Obtain the current voltage and current data of the transformer; Temperature distribution is predicted using the current voltage and current data and a pre-trained deep learning model to obtain the current prediction result of the transformer. The current prediction result includes at least the current winding short-circuit position, the current risk position, and the current temperature distribution of the transformer. The risk position is used to indicate the location where the transformer has a risk of winding burnout. The deep learning model is trained using a pre-collected training sample set, which includes at least the first risk position, the first voltage and current data, and the first temperature distribution under each winding short-circuit position of the transformer. The method further includes: Construct a finite element model of the transformer; An initial simulation is performed based on the finite element model to obtain the initial characteristic parameters of the finite element model, wherein the initial characteristic parameters include at least the maximum temperature parameter; Using each preset winding short-circuit position as an independent variable, the finite element model is used to simulate the winding short-circuit condition, and the first voltage and current data and the first temperature distribution under each winding short-circuit position are obtained. Based on the comparison between the maximum temperature parameter and the first temperature distribution, the first risk location under the short circuit location of each winding is obtained; The construction of the finite element model of the transformer includes: The finite element geometric model of the transformer is divided into the core, windings, and oil tank. The materials of the core, windings and tank are set according to the actual transformer parameters to obtain the set finite element geometric model; By setting the corresponding boundary conditions for the core and windings in the finite element geometric model, the finite element model of the transformer is obtained. The initial simulation based on the finite element model to obtain the initial characteristic parameters of the finite element model includes: The initial simulation of transient and steady-state simulation is performed by applying excitation to the finite element model to obtain the second current and voltage data of the transformer, and the maximum temperature parameter of the transformer during the transient and steady-state simulation is calculated. The step of simulating the short-circuit condition of the finite element model using each preset winding short-circuit position as an independent variable to obtain the first voltage and current data and the first temperature distribution at each winding short-circuit position includes: The short-circuit positions of the windings are traversed sequentially. For each target short-circuit position, the short-circuit condition of the transformer is set as the target short-circuit position to simulate the short-circuit condition. The first current and voltage data under the target short-circuit position are obtained, and the first temperature distribution of the transformer under the short-circuit condition at the target short-circuit position is calculated.

2. The method according to claim 1, characterized in that, The step of comparing the maximum temperature parameter with the first temperature distribution to obtain the first risk location under each winding short-circuit location includes: Each temperature data in the first temperature distribution is compared with the temperature threshold corresponding to the maximum temperature parameter to obtain the comparison result; If the comparison result is that the temperature data is greater than the temperature threshold, then it is determined that the transformer location corresponding to the temperature data is at risk of burning out, and is marked as the first risk location.

3. A deep learning-based transformer temperature distribution monitoring device, characterized in that, The device includes: Data acquisition module: used to acquire the current voltage and current data of the transformer; Temperature prediction module: used to predict the temperature distribution using the current voltage and current data and a pre-trained deep learning model, to obtain the current prediction result of the transformer. The current prediction result includes at least the current winding short-circuit position, the current risk position, and the current temperature distribution of the transformer. The risk position is used to indicate the location where the transformer has a risk of winding burnout. The deep learning model is trained using a pre-collected training sample set. The training sample set includes at least the first risk position, the first voltage and current data, and the first temperature distribution under each winding short-circuit position of the transformer. The device further includes: Model building module: used to construct the finite element model of the transformer; Parameter determination module: used to perform initial simulation based on the finite element model to obtain the initial characteristic parameters of the finite element model, wherein the initial characteristic parameters include at least the maximum temperature parameter; Winding short-circuit simulation module: used to simulate the short-circuit condition of the finite element model with each preset winding short-circuit position as the independent variable, and obtain the first voltage and current data and the first temperature distribution under each winding short-circuit position. Risk location module: used to compare the maximum temperature parameter with the first temperature distribution to obtain the first risk location under each winding short circuit position; Specifically, the model building module is used to divide the finite element geometric model of the transformer into the core, windings, and tank; to set the materials of the core, windings, and tank according to the actual transformer parameters to obtain the set finite element geometric model; and to set the corresponding boundary conditions for the core and windings in the set finite element geometric model to obtain the finite element model of the transformer. Specifically, the parameter determination module is used to apply excitation to the finite element model for the initial simulation of transient and steady-state simulation, obtain the second current and voltage data of the transformer, and calculate the maximum temperature parameter of the transformer during the transient and steady-state simulation. Specifically, the winding short-circuit simulation module is used to: sequentially traverse the winding short-circuit positions; for each target winding short-circuit position traversed, set the transformer's short-circuit condition as the target winding short-circuit position and perform short-circuit condition simulation; obtain the first current and voltage data under the target winding short-circuit position; and calculate the first temperature distribution of the transformer during operation under the short-circuit condition at the target winding short-circuit position.

4. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, the processor performs the steps of the method as described in any one of claims 1 to 2.

5. A computer device, comprising a memory and a processor, characterized in that, The memory stores a computer program that, when executed by the processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 2.

Images